- TopoJSON supports a very limited idea of a "folder": a feature or group of features can have a shared name which identifies them. For example, a dataset of country, state, and county borders can group the state borders under the name "state" and so on. This is, however, limited: there's no nesting and no additional attributes beyond the name.
- There's an additional specification that builds on GeoJSON to add styles: simplestyle-spec. Placemark can export simplestyle-spec encoded styles in GeoJSON, but the GeoJSON format itself doesn't specify how styles can be expressed in GeoJSON. So that one is a mixed answer: you can store styles in GeoJSON, but it's not guaranteed that every place you put that GeoJSON, it'll be styled.
A great default file format is GeoJSON. It's natively supported by map viewers like Mapbox GL, Leaflet, and even the Google Maps API. You can also analyze GeoJSON using desktop tools like QGIS and libraries like Turf.js.
You'll probably encounter a range of file formats, though - KML if you're using Google Earth, GPX and TCX if you're loading data from your GPS tracker, Shapefiles if you're working with legacy GIS systems, WKT if you're interacting often with databases like PostGIS. Placemark happily supports a wide range of geospatial formats, as do many other mapping tools.
Even though they're limited in the kinds of geometries they can reliably represent, CSV and Excel (XLS) tend to be very useful for collaborating with large groups with varied technical experience, because you can edit these formats in spreadsheet tools like Google Docs and Excel.
- GeoTIFF and EXIF are both examples of geospatial data stored in image formats, which explains how they're outliers. Placemark can import the data from both of them - the point at which an image is taken, in the case of EXIF, and the land area covered by a satellite image, in the case of GeoTIFF. But Placemark isn't an image editor, so it can't edit or export either EXIF or GeoTIFF data.
- WKT and Polyline are geometry encodings only, so they don't encode properties, folders, or anything other than geometry data. Usually when you're using one of them, the properties are stored elsewhere, like in other columns of a database.
- GeoJSONL, and FlatGeobuf are all approximately equal to the GeoJSON format, with slight changes. FlatGeobuf aims to be smaller than the equivalent GeoJSON by using techniques like quantization and, in the case of FlatGeobuf, a binary encoding. GeoJSONL is just GeoJSON but it splits each feature onto its own line so that very large files can be read efficiently.